Illustrating the behavior of the Badapple scoring function.
score = sActive / (sTested + median(sTested)) * aActive / (aTested + median(aTested)) * wActive / (wTested + median(wTested)) * 1e5
Where
sTested (substances tested) = # tested substances containing this scaffoldsActive (substances active) = # active substances containing this scaffoldaTested (assays tested) = # assays with tested compounds containing this scaffoldaActive (assays active) = # assays with active compounds containing this scaffoldwTested (samples tested) = # samples (wells) containing this scaffoldwActive (samples active) = # active samples (wells) containing this scaffoldNormalizing aggregate statistic parameters from PubChem assays (“Badapple: promiscuity patterns from noisy evidence”, Yang JJ, Ursu O, Lipinski CA, Sklar LA, Oprea TI Bologa CG, J. Cheminfo. 8:29 (2016), DOI: 10.1186/s13321-016-0137-3).
[1] "min_sTotal: 1" "max_sTotal: 35884"
[3] "min_sTested: 0" "med_sTested: 2"
[5] "max_sTested: 32346" "min_sActive: 0"
[7] "med_sActive: 1" "max_sActive: 20074"
[9] "min_aTested: 0" "med_aTested: 453"
[11] "max_aTested: 528" "min_aActive: 0"
[13] "med_aActive: 3" "max_aActive: 508"
[15] "min_wTested: 0" "med_wTested: 517"
[17] "max_wTested: 11603660" "min_wActive: 0"
[19] "med_wActive: 3" "max_wActive: 93128"